A Decoherence-Aware Quantum State and Liveness Detection Framework for robust live-spoof discrimination and identity-liveness dependency stability modeling improves robustness against identity-preserving attacks.
Abstract
Face anti-spoofing is essential for protecting biometric systems from presentation attacks such as print, replay, cut-photo, and deepfake manipulations. This work proposes a Decoherence-Aware Quantum State and Liveness Detection Framework for robust live-spoof discrimination. Facial video frames are first preprocessed and then passed through multi-scale convolutional networks to capture fine-grained spoof traces, such as moiré patterns, reflection noise, and display artifacts. The extracted features are then transformed into latent temporal states using superposition modeling and reversible temporal evolution. A stable evolution-adaptive transition mechanism is introduced to detect temporal disturbances induced by spoofing. Further, identity-liveness dependency stability modeling improves robustness against identity-preserving attacks. Quantum uncertainty-guided anomaly scoring is used for final spoof discrimination, followed by a lightweight MLP classifier. Experimental results on the CelebA-Spoof, OULU-NPU, and CASIA-FASD benchmark datasets demonstrate that the proposed framework achieves consistent performance across these evaluated datasets. The reported robustness and generalization are supported within the scope of these benchmark evaluations. On CelebA-Spoof, the model achieved APCER of 1.78%, BPCER of 2.29%, and ACER of 2.04%. On OULU-NPU, APCER, BPCER, and ACER were 1.38%, 1.42%, and 1.40%, respectively. On CASIA-FASD, the framework achieved 96.99% multiclass spoof classification accuracy, confirming better robustness and generalization.
Multi-modal anti-spoofing aims to differentiate live users from spoofing attacks using multiple biometric modalities during model training. While existing anti-spoofing methods often incorporate just one biometric modality, the effectiveness of attacking two or more biometric traits remains questionable. In this work, we introduce the multi-modal anti-spoofing approach to detect spoofing attacks across face and fingerprint. Our framework is built around an Angular Margin Loss (ArcFace) that increases interclass separation without disrupting cross-modal alignment, which enables reliable spoof detection across both face and fingerprint biometric characteristics. Moreover, to enhance model generalization against unseen spoof attacks, we include three adversarial attacks (i.e., FGSM, PGD, DeepFool) to evaluate our system. Extensive experiments on multi-modal benchmarks show that the proposed method not only significantly outperforms previous anti-spoofing methods but also uniquely offers the ability to handle potential attack types.
With the explosive growth of generative AI technologies, the threat posed by spoofing attacks and deepfake attacks on biometric authentication systems continues to grow. Traditional biometric systems rely on visual presentation and are subject to various types of presentation attacks such as replay attacks, printed photographs, and synthetic deepfake identities. This paper presents an Adaptive Multimodal Liveness Detection Framework (AMLF) to increase the strength of biometric identity systems and reduce their vulnerability to both spoofing and deepfake attacks. The framework also aims to reduce the computational costs of performing liveness detection on resource-constrained edge devices. The proposed framework uses three techniques for liveness detection across multiple biometric modalities (face, fingerprint, and iris): spatial texture analysis, temporal biometric signals, and frequency-domain artifacts. A lightweight, deep neural network architecture is used for the multimodal liveness detection process on edge devices and enables real-time liveness detection. A series of experiments conducted on publicly available datasets (FaceForensics++ , CASIA-Iris-V4, MSU-MFSD, and FVC2006) demonstrate that the proposed AMLF considerably increases the accuracy of detecting spoofing attacks over existing approaches based on deep learning methods while significantly reducing the computational costs associated with these methods. Overall, the AMLF framework provides a highly effective means to improve the capability of next generation biometric authentication systems, particularly in an edge computing environment.
Ankita Kotalwar, R. Joshi· Discover Artificial Intellig...· 0 citations
A multi-dimensional feature fusion-based face anti-spoofing detection method based on the YOLOv8 architecture that integrates dynamic optical flow features with static texture analysis and achieves robust detection through the fusion of spatial and temporal cues.
Yanhua Liang, Pengcheng Zhou, Hongmei Qin et al.· International Conference on...· 0 citations
Identity verification is a prevalent application of facial recognition systems, but the traditional feature embeddings used in the systems create biometric data vulnerable to many privacy violations. Traditional privacy-preserving solutions are typically either recognition-accuracy compromised or have high computational costs, making them ineffective in real, resource-bounded world settings. Although the current classical methods cannot offer intrinsic template protection without affecting the performance, the current hybrid quantum-classical methods do not have an end-to-end deployment-friendly architecture that has been demonstrated to be reliable in facial recognition in the presence of realistic adversarial and inversion attacks. To solve these problems, a hybrid classical-quantum processing pipeline is suggested. Here, lightweight convolutional feature extraction is done fully on-device, followed by the encoding of the normalized embeddings into quantum states and conversion by a variational quantum circuit (VQC). Privacy is physically enforced, so that only non-invertible measurements of quantum measurements are sent to infer classically. Experimental analyses prove the suggested model to be significantly better than lightweight and privacy-oriented baselines. In particular, LFW and CelebA accuracies reach 94.3% and 92.8%, respectively, with zero template recoverability and inference latency of less than 200 ms, which proves the strength and efficiency of the system in the edge applications, as well as the guidelines for prospective authors who will have to prepare the final manuscript accepted for publication.
Farah Saad Al-Mukhtar, Rana M. Hasan, Raghad AbdulHadi AbdulQader· Basrah journal of science· 0 citations
In the contemporary digital landscape, the exponential proliferation of high-dimensional multimedia data across social platforms, communication networks, and biometric authentication channels is accompanied by an escalating threat of sophisticated generative deception. Deepfakes and synthetic media manipulations present critical systemic risks, ranging from targeted identity fraud to widespread misinformation campaigns. Traditional forensic methodologies—such as pixel-level error level analysis, lighting inconsistency checks, and static rule-based verification—fail to scale efficiently against modern deep synthesis techniques due to heavy compression assumptions, manual feature-engineering constraints, and computational latency. To address these challenges, this monograph presents the design and deployment of the Deepfake Forensic Suite, an automated Identity Mapping and Media Integrity Verification System. The proposed framework establishes a multi-layered security pipeline. First, it implements a high-precision biometric mapping and alignment phase utilizing Multi-task Cascaded Convolutional Networks (MTCNN) to isolate facial regions and eliminate environmental noise. Second, it leverages an optimized MobileNetV2 architecture to extract deep spatial features and compress complex visual attributes into a compact latent representation. By learning the structural characteristics of authentic human faces, the system computes principled prediction probability scores that naturally diverge when processing synthetic manipulations. Furthermore, a statistically robust tri-state classification strategy (Real, Fake, or Uncertain) is established based on validation-set confidence percentiles, enhancing forensic reliability by flagging borderline cases for manual administrative review. The performance of the system is evaluated against established baselines, including traditional Viola-Jones frameworks and shallow convolutional structures. Finally, the practical deployment-readiness of the system is demonstrated through model serialization, a real-time webcam inference API, and a reproducible, interactive web dashboard engineered entirely within the Streamlit framework. The resulting suite provides a lightweight, high-assurance digital forensics solution capable of edge-device execution without requiring slow, cloud-dependent infrastructure.
T. Manimala, P. Sravani, V. Rajitha et al.· EPRA international journal o...· 0 citations
Face recognition systems are vulnerable to presentation attacks including printed photographs, mobile-screen replay, and video replay. Existing multi-modal anti-spoofing systems apply fixed branch weights during score fusion, which degrade under varying environmental conditions such as low illumination, motion blur, or extreme head pose. We propose GazeHawkAI, a three-branch anti-spoofing framework incorporating a novel Context-Aware Entropy-Guided Adaptive Fusion (CEAF) layer that dynamically re-weights branch contributions using Shannon entropy of per-branch score histories and real-time environmental quality metrics. Branch A performs ArcFace identity verification; Branch B analyses eye dynamics through Eye Aspect Ratio (EAR), blink detection, and a randomised two-step gaze challenge; Branch D evaluates face depth and texture via MediaPipe landmark geometry, Local Binary Pattern (LBP) entropy, Fast Fourier Transform (FFT) moiré detection, and a temporal video-replay detector based on pixel-variance decomposition. The CEAF layer computes per-branch confidence as Qi×Ci, normalises weights with a minimum floor of 0.05, and feeds the result to a weighted geometric-mean fusion. Benchmark testing on NUAA, Replay-Attack, and MSU-MFSD yields HTER 12.1% on Replay-Attack. This work additionally validates on a self-collected dataset of N=21,239 samples spanning five attack categories, where CEAF reduces ACER from 45.07% to 39.63% (a 5.44 percentage-point improvement over fixed-weight fusion) and AUC from 0.585 to 0.629. The system runs in real time on a CPU-only laptop with no cloud dependency.
Jagmit Singh, D. Makhija, H. Patil et al.· 2026 7th International Confe...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
